Neuroimaging in the Definition and Organization of the Epilepsies: We’re Not There Yet

SUMMARY

Neuroimaging significantly impacts the diagnosis and treatment of epilepsy patients. Despite its importance, MRI has been marginally incorporated into concepts used to define epilepsy etiologies by the ILAE Classification Commission. We propose that Structural etiology be defined as positive neuroimaging abnormalities likely causing the seizures. This would contrast with Genetic and Unknown etiologies where imaging shows no overt structural abnormality that explains the seizures. It is further recommended that Structural and Metabolic be separated into individual categories as the outcomes and therapies are different. It is advocated that Structural etiology be subdivided into subgroups based on MRI and surgical syndromes. With this approach, the ILAE should acknowledge that both MRI and EEG are necessary diagnostic tools in the classification of epilepsy syndromes and etiologies in the modern era. Promoting the use of neuroimaging into concepts that determine terminology will promote the notion that epilepsy classification should consider structural etiology of the seizures, along with the frequency of the most common epilepsy syndromes, and prognosis for spontaneous and treated remission and cure.

Introduction

Over the past 35 years, neuroimaging, especially MRI, FDG-PET, and ictal SPECT, has had a profound impact on the diagnosis and management of patients with epilepsy, and epilepsy neurosurgery. Epidemiology studies indicate that lesions likely responsible for the seizures are found in 20% to 25% of MRI scans from patients with non-idiopathic (non-genetic) epilepsy syndromes (Berg et al. 2009b). Children with positive MRI scans have a younger age at seizure onset, and are more likely to meet the definition of medically refractory epilepsy than those with negative scans (Berg et al., 2009b; Harvey et al., 2008; Kwan et al. 2010). Once a patient has failed adequate trials of two anti-epilepsy drugs (AED), the chance of becoming seizure free with AEDs falls to under 5% for those with positive MRI scans compared with higher rates for patients with negative scans (Berg et al. 2009a; Callaghan et al. 2011). Over 90% of epilepsy surgery patients with medically refractory epilepsy have positive neuroimaging (Harvey et al. 2008). In MRI negative cases, newer imaging modalities, such as FDG-PET co-registration and SISCOM are identifying occult structural lesions, and complete resection of the lesion is a major predictor of becoming seizure free after surgery (McIntosh et al. 2004; Salamon et al. 2008; Velasco et al. 2011; von Oertzen et al. 2011; Wu et al. 2011).

Despite the documented importance of neuroimaging in the care of patients with epilepsy, MRI and other neuroimaging tools have played a fairly minor role in the conceptual organization and definition of terms used for classifying the epilepsies by etiology. This report reviews the role of neuroimaging in past and current classification schemes proposed by the International League Against Epilepsy (ILAE). Based on that history and the importance of MRI in current practice, we propose changes that better integrate neuroimaging into the organization of the epilepsies. It is our position that in the modern era MRI should be equally available and the findings given equal weight compared with electroencephalography (EEG) and seizure semiology in the organization of epilepsy syndromes and etiology.

Neuroimaging in the ILAE Classification of Epilepsies and Etiology

The ILAE classification of the epilepsies and their subsequent revisions have focused almost exclusively on behavioral semiology and electrophysiological characteristics with minimal input from neuroimaging in defining epilepsy syndromes (Commission on Classification and Terminology of the International League Against Epilepsy, 1981; 1985; 1989). This is expected, since the initial concepts that created these schemes were produced before the widespread availability of modern neuroimaging (Gastaut 1969b; Gastaut 1969a). Consequently, terms used to define etiology have been impractical for those conducting clinical research at tertiary specialty centers that include neuroimaging and epilepsy neurosurgery. In the 1985 report and 1989 revision, for example, the concept of partial epilepsies (now termed focal) were presumed to be from known or suspected central nervous system (CNS) insults and abnormalities, and were considered symptomatic. The notion was that seizures were secondary in symptomatic epilepsy as compared with primary epilepsy for those with traditional idiopathic etiologies. Despite the availability of CT and MRI, there was no recommendation by the Commission to confirm the existence of structural lesions in those with symptomatic epilepsy. These limitations to the ILAE classification system were recognized more than a decade ago (Engel 1998; Everitt and Sander 1999).

The logic, while reasonable in concept, has poorly correlated with MRI findings in patients with epilepsy. For example, it is estimated that 10% to 15% of patients with “symptomatic epilepsy” have negative MRI scans (Berg et al. 2009b). In addition, some patients with traditional cryptogenic and idiopathic epilepsy using electro-clinical criteria often harbored lesions related to their epilepsy (Loddenkemper et al. 2009). This has generated considerable confusion in attempting to identify epilepsy patients with structural lesions as the etiology of their seizures. In the 1989 classification system, patients were considered to have symptomatic etiology if there was a prior history of a presumed CNS insult (e.g. meningitis, trauma) and focal seizures. That patient was still considered to have symptomatic epilepsy even if the MRI and neurological exam was normal. Further, it was impossible to separate a patient with the same clinical history but with focal damage to the anterior temporal region rendering that person a possible surgical candidate. Similarly, if patients developed recurrent focal epilepsy without a known prior CNS insult their epilepsy was generally classified as cryptogenic until the MRI showed a lesion and then they might be re-classified as symptomatic (Shinnar et al. 1999). Furthermore, some patients with generalized epilepsy and seizures starting as infants, such as infantile spasms and Lennox-Gastaut syndrome, were found to have focal cortical lesions by neuroimaging that when surgically removed controlled their seizures (Chugani et al. 1990; Wyllie et al. 1996; Nordli et al. 2001). These findings challenge the central concept of the ILAE classification systems that separates primary from secondary epilepsy.

Proposed Amendments to the 2010 Commission on Classification Report

The 2010 report from the Classification Commission, while better, does not go far enough in incorporating MRI and other neuroimaging techniques into definitions of epilepsy etiology (Berg et al. 2010). In fact, there were essentially minimal changes in the 2010 report to what constitutes the concept of structural compared with symptomatic etiology. As a result, modifications are suggested that better reflect the role and importance of neuroimaging in current clinical practice that has impact for patients. For instance, the 2010 report combined Metabolic with Structural into a single etiology category. This was done to emphasize the difference of genetic from non-genetic etiologies, as the committee focused considerable attention on the future contribution of gene research in identifying epilepsy syndromes. However, that discussion did not consider the current contribution of neuroimaging in determining structural epilepsy etiologies. Grouping metabolic syndromes with structural lesions into a single etiology group does not make sense for practicing epileptologists as the syndromes and treatments are different. We recommend that Structural and Metabolic etiologies be independent categories in any classification of epilepsy etiology (Table 1).

Table 1.

Recommended Amendments to the 2010 Report from Classification Commission

Amendments to Etiology Classification

Acknowledge that EEG and MRI are necessary diagnostic tools to classify epilepsy by etiology Structural and Metabolic should be separate etiology categories (not combined) Structural Etiology should be defined as a lesion on MRI (and possibly other neuroimaging     modality) likely causing the seizures Within Structural Etiology, subgroup most common etiologies by MRI features Within Structural Etiology, capture common and rarer epilepsy surgery syndromes under     separate sub-group, which incorporates negative MRI but positive histopathology Unknown etiology should be defined as no known electro-clinical syndrome by EEG and no     identified etiology by MRI Genetic etiology should include in the definition that there is no apparent structural lesion     responsible for the epilepsy

Another problem with the 2010 report is that a positive MRI was not considered essential and necessary in the definition of epilepsy patients with Structural etiology. In clinical practice, MRI is used just as often if not more frequently than EEG as diagnostic tools in patients with recurrent seizures. In fact, in many parts of the world and in rural America, MRI is more widely available than competent EEG evaluation. Given that one-fifth to one-quarter of epilepsy patients are MRI positive and if positive such information helps predict response to medical treatment and prognosis, it seems essential that MRI be given equal weight to EEG in the classification of epilepsy etiologies. Hence, it is strongly recommended that classification by etiology recognize the importance of obtaining brain MRI, and that by definition all patients classified as Structural have positive scans related to their seizures. This is irrespective of the seizure type as it has become increasingly recognized that patients with discrete brain lesions can have different epilepsy phenotypes often depending on age at seizure onset.

Even with the above proposed changes, it would still be difficult to use the 2010 ILAE report in capturing epilepsy surgery syndromes and their histopathologic etiologies. Given the importance of identifying these patients and the lack of appreciation among general physicians and neurologists about treatment options in those with positive MRI scans, this should be changed (Erba et al. 2012). While one could argue that epilepsy surgery represents a relatively small proportion of patients with epilepsy (roughly 5% of those with new onset seizures), of those with medically refractory epilepsy at a frequency of one seizure per month or greater (about 15% to 20% of those with epilepsy) from 1:3 to 1:4 will be surgical candidates (Berg et al. 2009b). This is similar in frequency as patients with absence epilepsy and benign epilepsy from central-temporal spikes (Wirrell et al. 2011). Likewise, identifying and characterizing the incidence and prevalence of potential epilepsy surgical cases has an impact on public health policy and planning health care resources, such as the number and staffing requirements of comprehensive epilepsy centers.

For these reasons, it is urged that the Structural category be organized in a way that captures the most frequent etiologies by MRI characteristics with separate sub-groups for the most common epilepsy surgery syndromes and MRI negative but neuroimaging positive (e.g. other modalities such as FDG-PET, SISCOM, MEG) cases (Table 2). The epilepsy surgery sub-category can incorporate histopathology along with neuroimaging into the classification by etiology, which would help capture subtle substrates into the Structural etiology group. This organization plan anticipates that knowledge gained from newer neuroimaging techniques, such as resting state fMRI, can be more easily incorporated into concepts of etiology (Zhang et al. 2011; Masterton et al. 2012). From existing epidemiology studies, the most common structural etiologies described by MRI are likely acquired associated with atrophic brain lesions (focal or generalized brain atrophy secondary to various insults such as trauma, stroke, birth injury and hemorrhage, porencephalic cysts, infection, etc.), and malformations of cortical development (e.g. lissencephaly, polymicrogyria, microcephaly, etc.) (Barkovich et al. 2005). Less common are structural lesions secondary to genetic mutations (e.g. tubers in TSC, lesions and tumors from neurofibromatosis), and discrete lesions (e.g. low-grade tumors, hypothalamic hamartoma. For epilepsy surgery syndromes, patients with mesial temporal lobe epilepsy from hippocampal sclerosis, seizures from tumors (temporal and extratemporal), and refractory epilepsy from Type I and Type II focal cortical dysplasia (FDC) are the most common etiologies (Harvey et al. 2008; Hemb et al. 2010; Blumcke et al. 2011). Rarer syndromes will include a number of etiologies as suggested in Table 2.

Table 2.

Proposed Classification by Etiology incorporating MRI

Structural Etiology: Positive MRI lesion likely causing epilepsy	Approximate Incidence
1.0 Genetic Primary (Fits electro-clinical syndrome or genetic    mutation; generally negative MRI scan)
2.0 Structural Primary
2.1 MRI Description
Atrophic/Acquired secondary to injury (e.g. trauma)	40–50%
Malformation of Cortical Development	20–25%
Discrete Lesions (e.g. tumors, caveromas)	15%
Lesions from Genetic Mutations (e.g. TSC, NF1)	10%
Others not described above
2.2 Epilepsy Surgery Syndromes (confirmed by histopathology   when possible)
MTLE from HS	30–40% Adults
Tumors-operated	20% Adults and Children
Focal Cortical Dysplasia	30–40% Children
Histopathology Negative
Rarer Epilepsy Surgery Syndromes (examples)	All less than 5%
Hypothalamic hamartoma
Rasmussen Encephalitis
Tuberous sclerosis complex
Hemimegalencephaly
Sturge-Weber
Cavernous malformations
Aircardi Syndrome
Hypomelanosis of Ito
Otahara Syndrome
2.3 MRI negative but neuroimaging positive lesions
3.0 Unknown (negative MRI and no identified electro-clinical    syndrome)
Focal Seizures
Temporal	To be determined
Frontal
Parietal
Occipital
Generalized non-electro-clinical syndrome

Implications of Suggested Changes on Clinical Research and Coding

The inclusion of neuroimaging into the concepts and terminology defining epilepsy etiology has other implications that should be considered by the Commission on Classification and clinical researchers. For example, while it is implied by the way epilepsy syndromes are described that EEG capability is necessary, the 1989 and 2010 Commission on Classification reports do not explicitly state what are the minimal requirements of expertise and diagnostic tools necessary for classifying patients with epilepsy. Clinical criteria alone are usually not enough to fully characterize a person’s seizures. Without a witness to the event, the best that probably can be cataloged is whether a person has epilepsy, probable epilepsy, or a paroxysmal disorder of unclear etiology (Thurman et al. 2011). If one accepts that characterizing seizures and epilepsy syndromes requires more than second hand clinical information on semiology, then it is suggested that requirements to use the ILAE classification system be incorporated into the report, and EEG and MRI be included as minimal criteria. An extension of that concept has impact on defining patients placed in the Genetic and Unknown categories. To diagnose someone as having epilepsy of Unknown etiology takes on a more refined definition as a person who has a negative MRI scan and an unidentified electro-clinical syndrome (Linehan et al. 2011). Currently, Unknown etiology is not well defined in the 2010 report.

Incorporation of MRI into the definition of etiology has impact on epidemiology studies and the World Health Organization’s International Classification of Diseases (WHO-ICD). With few exceptions, nearly all epidemiology studies reported to date and the ICD-9 and ICD-10 codes have not incorporated MRI scans and their results. MRI findings are used in other neurological classification systems, such as tumors and vascular malformations. Inclusion of MRI in epilepsy diagnosis for population surveys and hospital reporting would provide critical information for clinicians and health care policy makers about the incidence and prevalence of epilepsy patients by specific etiologies.

ILAE Definition of Medically Refractory Epilepsy and Etiology

As previously stated, patients with positive MRI scans are less likely to be controlled on AEDs than those with normal scans. In the future, it might be worth considering inclusion of ILAE’s definition of therapy resistant epilepsy and therapeutic response to medical and surgical treatments as concepts applied in classifying seizures and epilepsy (Kwan et al. 2010). At the end of the day, what patients most want to know is whether there is a treatment for their particular form of epilepsy and what are the chances that therapy will work? If existing treatments, such as currently available AEDs, have the same response regardless of the type of epilepsy or seizure type the value of that classification system should probably be called into suspect, and another scheme with greater focus on incidence and prognosis considered (Glauser et al. 2006).

Future Considerations: Etiology versus Causality

The 2010 Commission on Classification report starts the process of organizing epilepsy syndromes as genetic and structural in origin while still recognizing that etiology is unknown in about half of patients with epilepsy (Berg et al. 2010). The report also acknowledges that any classification should be flexible and change as new information becomes available and as problems with the older system are encountered. A long interval passed from 1989 to 2010 between the last two major reports and it is hoped that further refinements from the ILAE Commission on Classification appear sooner than later. This is especially relevant for neuroimaging where continued improvements and upgrades can be expected to have clinical impact on patients with epilepsy (Hemb et al. 2010).

As our diagnostic tools improve, we also need to keep in mind that the etiology of a person’s epilepsy is not always the same as the “cause” of the seizures. This concept may need to be considered as future terms in epilepsy are defined. A patient may have hippocampal sclerosis or cortical dysplasia as the etiology of their epilepsy, but that does not mean we know how the lesion produces seizures. The lesion may alter excitatory and inhibitory synaptic transmission or alter hard wired axonal circuits between the lesion and surrounding cortex. Understanding how genetic and structural abnormalities “cause” the brain to have intermittent unprovoked seizures will have greater impact on how we diagnosis and treat patients than just knowing etiology.

Perhaps future classification systems will incorporate a hybrid of existing concepts and methods of diagnosis as more fundamental “causes” of epilepsy are discovered. Such concepts will also have to take into consideration that epilepsy syndromes and networks are probably not static but evolve over the lifetime of a patient. This means understanding how seizures and epilepsy alter brain networks during development, and in matured brains. In other words, the classification of the epilepsies, epilepsy syndromes, and epilepsy etiologies should rely on all available diagnostic methods and concepts, especially if the findings from those studies have bearing on prognosis and treatment of the seizures. Presently, that should include clinical semiology, EEG, and where appropriate MRI and other imaging studies.